v0.8.4 Stable Release

Debug build errors with AI clarity

vpipe captures your CLI output, scrubs sensitive keys, and provides instant root-cause analysis. Stop scrolling, start shipping.

bash — vpipe-session
$ npm run build | vpipe
[vpipe] Initializing Kinetic Pipeline...
[vpipe] Intercepting stderr...
Error: AWS_ACCESS_KEY_ID is missing or invalid.
at S3Client.send (node_modules/@aws-sdk/...)
at deploy (scripts/deploy.js:42:12)
auto_awesome AI Analysis
"The build is failing because your .env.production file is not being loaded by the CI environment. The S3 client cannot authenticate."

The Kinetic Pipeline

Five steps from chaos to resolution.

input

Capture

Real-time interception of stdout and stderr from any CLI tool.

security

Sanitize

In-memory redaction of PII, keys, and environment secrets.

compress

Truncate

Smart filtering of repetitive logs to focus on the error signal.

psychology

Analyze

High-speed reasoning via weighted context algorithms.

check_circle

Resolve

Actionable fixes and automated PR suggestions.

Zero-Trust Sanitization

Data security is our primary directive. vpipe uses a high-performance regex engine and pre-trained NER models to ensure that no sensitive data ever reaches the AI providers.

  • verified_user

    Local Redaction

    All processing happens in your memory buffer. No raw logs are ever stored.

  • regular_expression

    Custom Pattern Matching

    Define your own internal security patterns via `.vpiperc`.

Raw Output (Unsafe)
vpipe Sanitized
POST /v1/deploy
Host: internal-api.prod
X-API-KEY: AKIA23528FJSK
DB_URL: postgres://admin:P@ssw0rd123@...
Status: 403 Forbidden
POST /v1/deploy
Host: internal-api.prod
X-API-KEY: [REDACTED_AWS_KEY]
DB_URL: [REDACTED_DB_URI]
Status: 403 Forbidden

Weighted Context Algorithm

AI models have token limits. vpipe maximizes efficiency by calculating an 'Error Signal Density' map across your logs.

Log Start (0ms) Primary Error Peak (Critical Context) Log End (12s)
100k+

Lines Processed

98.2%

Noise Reduction

<45ms

Analysis Latency

Connect to Your Intelligent Engine

Groq Logo ULTRA LOW LATENCY
OpenAI Logo GPT-4O SUPPORT
Anthropic Logo CLAUDE 3.5 SONNET
Mistral Logo OPEN WEIGHTS

vpipe OSS

The local-first developer companion.

MIT LICENSE
  • check Standard Sanitization Engine
  • check Custom LLM Endpoint Support
  • check Community Regex Patterns
Recommended

vpipe Cloud

Orchestrated intelligence for teams.

MANAGED
  • bolt Centralized Error Analytics Dashboard
  • bolt SOC2 Compliant Sanitization
  • bolt GitHub Action & Jenkins Plugins

Technical Q&A

How does vpipe protect my source code?

vpipe only captures the specific stack trace and error message identified by our signal detection. We do not upload entire source files. For Enterprise users, we offer a dedicated VPC deployment where no data leaves your network.

Can I use vpipe with non-English logs?

Yes. While our primary reasoning models are optimized for English-language error messages, our sanitization engine supports UTF-8 and multi-lingual NER (Named Entity Recognition) to protect data across global log sets.

What is the performance overhead?

Negligible. vpipe's interception engine is written in Rust and operates as a non-blocking stream. It adds less than 2ms of latency to your standard build process until an error is actually detected.

Does it work with Windows CMD/PowerShell?

vpipe is cross-platform. We provide native binaries for macOS (Intel/M1), Linux (x64/ARM), and Windows. Integration works seamlessly with bash, zsh, fish, and PowerShell.